The paper introduces PAI‑Bench, a benchmark designed to evaluate persistent AI agents on how faithfully they adhere to a versioned identity contract. It separates several dimensions—recall, composition, behavioral enactment, resistance, persistence, lineage, and role‑conditioned updates—while keeping scoring oracles independent of the target process. Experiments on synthetic profiles show that explicit cues can significantly alter the presence of identity identifiers, revealing prompt‑dependent component selection and sensitivity to startup cues.
arXiv:2608. 13482v1 Announce Type: cross Abstract: As language-model-based AI is increasingly deployed in autonomous settings, aligning its goals and values with those of humans becomes critical.
The paper introduces the Situated Identity Test (SIT), a framework that assesses whether a language model’s behavior can be traced to a specific developmental lineage rather than merely imitating a persona. SIT requires agents to possess accurate knowledge of their recorded experiences while appropriately ignoring ungrounded information, and it demonstrates that policies based only on compressed profiles are limited in distinguishing between colliding life histories. The authors present SITBench, an evaluation suite with 25 profile‑collision pairs and 10,000 probes across nine model architectures, and provide open‑source tools and pilot results on state‑of‑the‑art foundation models.
The paper investigates how personalized agents decide to use, ignore, update, or query retrieved user memory before acting on a task. An empirical audit protocol is developed to test structured intermediate outputs, revealing that while exposing state definitions improves accuracy, an explicit state-output field does not significantly enhance policy accuracy for large language models. The study also shows that example-level accuracy overstates consistency, with full four‑way family success being rare, and that providing benchmark‑associated state labels merely conditions predictions rather than proving internal fidelity.
arXiv:2608.31142v1 Announce Type: cross Abstract: The 2025--2026 AI market has seen a wave of stealth releases: frontier models launched anonymously on developer platforms under codenames. For their...
The paper introduces CELLAUDIT, a method for auditing whether inputs claimed to influence predictive models actually do so. By testing if an input can enter the computation, whether predictions depend on it, and if that dependence improves observed responses, the authors evaluate agent-generated predictors on a morphology‑transcriptomics benchmark (BBBC047). Their findings show that many models claim compound contributions that are not supported by the data, and that falsification‑guided revisions can recover genuine input effects while improving performance.
arXiv:2609.39882v1 Announce Type: new Abstract: Pre-training equips large language models (LLMs) with a broad repertoire of behavioral patterns associated with roles, styles, values, and goals. Post-...
arXiv:2608. 06485v1 Announce Type: cross Abstract: Personality-conditioned LLM agents (PC-Agents) are increasingly used in emotional support, social simulation, and role-playing, motivating the development of lifelong agents that remain coherent over extended interactions.
ContextEcho is a benchmark and harness designed to measure persona drift in large language models during long, tool‑using coding sessions. It includes a 25‑probe identity suite, a snapshot‑then‑probe protocol that preserves the main conversation, and both judged and judge‑free measurement surfaces. Across 23 frontier models and thousands of turns, the benchmark shows that persona drift is widespread, not limited to specific model families, and that simple in‑session compaction does not reset it, while a single‑shot anchor can restore the intended persona.
arXiv:2607. 14275v1 Announce Type: new Abstract: Context engineering has become central to building reliable AI agents, yet it remains largely unmeasured.
arXiv:2607. 28818v1 Announce Type: new Abstract: As AI companions increasingly mediate repeated social interaction, users may rely on a stable role and shared history, yet locally acceptable replies do not ensure that either persists.
The paper introduces the concept of perfect aliasing, where a truth probe that aligns truthful reporting with a task’s prescribed action cannot differentiate between the two based solely on its labels. In a binary reporting game, probes fitted on compliant contexts yield identical optimizations, while on rival contexts their labels are complementary, causing their AUROCs to sum to one across 751 cell-layer pairs. By employing randomized codebooks and mixed-context fitting, the authors demonstrate that separating prescribed output symbols from semantic action enables perfect recovery of truth, achieving an AUROC of 1.000 on rival trials for a reward-trained Gemma-2-9B policy, whereas conventional probes perform near chance.
arXiv:2605. 28969v2 Announce Type: replace-cross Abstract: If an AI agent makes decisions on a person's behalf, those decisions must align with its user.